[Bioperl-l] SearchIO speed up

Chris Fields cjfields at uiuc.edu
Mon Aug 14 12:32:36 EDT 2006


On Aug 14, 2006, at 11:00 AM, aaron.j.mackey at gsk.com wrote:

>>> Otherwise I could see an experimental branch stagnating over time
>>> without bug fixes, code updates, etc.
>>
>> Yeah, after a while once you've convinced yourself your
>> 'experimental' changes work, you should port them over to the main
>> trunk and then the experimental one should go into oblivion.
>
> To reemphasize the point, there's no reason why there should be  
> only one
> experimental branch; rather, when a (named) experiment is to be  
> performed,
> you branch to perform that experiment, the endpoint of which is either
> experimental success (merge back to trunk) or failure (branch is  
> relegated
> to oblivion).  The best way to ensure that your experimental branch is
> reflective of the current trunk is to make new experimental  
> branches from
> the trunk whenever you need one.
>
> Don't be afraid of branching, that's what it's there for!  Branch  
> early,
> branch often; don't pollute the trunk!
>
> -Aaron

Yeah,  I agree.  Any of Sendu's commits to the current experimental  
branch would eventually merge back to trunk once they prove  
successful and work on that experimental branch would stop.  Seems  
using branches for experimental code hasn't been taken advantage of  
nearly enough judging by Hilmar's statement about Bio::SeqFeatureI (I  
guess re: the original 1.5 release).

BTW, I really like the 'lazy parsing' used for SwissKnife and the use  
of seek() processing chunks.  This is definitely something to think  
about in the future for Bio::SeqIO.  Would be nice to have the  
capability of handling and writing very large sequences w/o memory  
issues.

Christopher Fields
Postdoctoral Researcher
Lab of Dr. Robert Switzer
Dept of Biochemistry
University of Illinois Urbana-Champaign





More information about the Bioperl-l mailing list